Almost half of new cancer diagnoses in adults in the U.S. will be of three types: cancers of the lung, breast, and prostate. If significant progress is to be made in reducing total cancer incidence, it will be important to finds means of preventing these major tumors. Toward this end, basic scientists have identified a variety of genes in experimental systems that may be important in the etiology of these cancers. Concordantly, epidemiologists have identified environmental factors that are associated with increased cancer risk (e.g., smoking and lung cancer). In addition, statistical geneticists have shown that for a large number of malignancies, the risk of developing and surviving cancer is not uniformly distributed throughout the population. Given the complexity of cancer phenotypes and the dramatic disruption of cancer cells' genomic constitution, it is rational to only be cautiously optimistic about the near term success of novel therapeutic approaches to end-stage disease. A more efficacious short-term strategy may be early diagnosis or prevention. For most cancer types, therapeutic intervention in early stages of disease has demonstrated high rates of efficacy. The laboratory is attempting to identify genetic constitutional markers that distinguish individuals at high risk for developing breast cancer, or at risk of developing complications from prevention interventions. Two sample populations are being used to achieve these goals: a case only collection from a large phase 3 trial of Tamoxifen (13,388 women) and a case-control study (750 in all). Preliminary studies in the laboratory using a distinct case-control population set (493 in all) have focused on candidate breast cancer susceptibility genes, including those involved in estrogen biosynthesis and metabolism. A significant difference in allelic distribution of the aromatase gene (Cyp19) was observed between cases and controls. In addition, the laboratory has developed MALDI-TOF MS assays on an extended collection of gene markers useful in dissecting estrogen biosynthetic and metabolic pathways, focusing specifically on functionally significant allelic variants described in the literature. Thirty-eight variants are being assessed in the case only Tamoxifen studies to address potential associations with complications from prevention intervention, and in the case-control population to identify markers indicative of high risk for breast cancer development. The laboratory is also attempting to identify genes that modify the risk of developing lung cancer by examining candidate gene variation in a case-control study. In this study, fifteen variants in fourteen genes involved in Phase I and Phase II metabolism were evaluated. When these 15 candidate susceptibility gene variants were assessed in a 756-sample subset of smokers, and the results introduced into logistic regression models (following adjustments for genetic background), only GSTA4 was significantly associated with lung cancer. The interaction among tobacco smoke, dietary intake and candidate gene variation was also determined. On stratified analysis of 329 smokers, a healthy dietary pattern (high intakes of fibers and carbohydrates and low intakes of protein and animal fat) was associated with decreased lung cancer risk among GSTM1 null individuals. This result suggests that dietary factors can influence carcinogen metabolizing enzymes effects in lung cancer risk. Pilot application of path analysis methodologies for interpreting lung cancer gene expression has been performed utilizing the G1/S checkpoint pathway. This pathway was chosen because it has been shown to play an important role in the development of lung cancer. The initial study examined whether the fitted models of correlation between genes reflect current biological knowledge about the changes that occur in this pathway in lung tumors. Gene expression data from 17 normal lung and 55 lung carcinoma cell lines was used for path analysis. Much is known about many of the key genes in the pathway, which includes TGF-b, SMAD4, p53, gsk3b, cdc25A, p27, p21, cdk2, cyclin E, cdk6 cdk4, cyclin D and Rb, and models were tested based on the available data. The first and simplest model examined the direct influence of upstream genes on Rb expression. In normal cells, 35% of the variation in Rb expression appeared related to the expression of cdks and cyclins. Increases in these proteins, which signal G1 to S transition, corresponded to decreases in Rb expression. Tumor samples showed lower Rb expression, and Rb expression was not strongly influenced by cdk4 and cdk6. The influence of cyclin E over Rb expression appeared to increase in tumors, but there was an apparent decoupling of cyclin E and its regulator cdk2 suggesting that these effects may be poorly regulated. This initial study is consistent with the literature concerning these genes in normal and cancer cells, and suggests that path analysis can be used to reliably test hypotheses concerning changes in the patterns of correlations occurring in the normal and disease state.